Managing cold data. Amazon Redshift is a cloud-native data warehouse from AWS that is petabyte scalable. A petabyte is 1000 terabytes and Redshift can store up to two petabytes of raw data. There are a pair of issues that come with this much data storage. One of them is conceptualizing this much data. The other, and probably a more practical problem, is dealing with cold data. That is, data that is infrequently accessed. 

One way to deal with cold data is to simply delete it. Problem solved, right? Well, maybe. But if that data is needed, even once or twice a year, deleting it from the cluster will create a new problem. How will you restore it when you need it? Keeping it in the cluster, however, is expensive. I'm not sure that I know of anyone who wants to pay for cluster space for data that is rarely used and grows in size every year. To address this issue, AWS launched Redshift Spectrum in 2017.

Spectrum, as part of Amazon Redshift, uses SQL to query data that is stored as files in Amazon S3. More than that, it can do SQL joins between S3 data and tables on the Redshift cluster. It's hot and cold data working seamlessly together. "Any sufficiently advanced technology is indistinguishable from magic," Arthur C. Clarke. To me, this is nothing short of magical. By putting data in S3 and treating it like a native Redshift table, it saves space on Redshift clusters and reduces storage costs. As a bonus, with more available space and less overhead regarding data distribution and networking, overall query performance is increased.